AMD needs to do something for a refresh because TMSC royally screwed up with the next process node. So rather it being a revolutionary step it has become an evolutionary one. If that makes any sense.
Printable View
GPU-Z SPs error or wavefront is still 80 (20 x 4) :confused:
http://www.dinoxpc.com/public/images...c55f3_6900.jpg
5870 (fillrates):
http://www.geeks3d.com/public/jegx/2...5870-gpu-z.jpg
AMD's own Fermi launch. Yikes pls tell me this isn't true.
GPU-Z is just a database, it doesn't know its actual units or any of that stuff. It can tell you clocks and that's about it - the rest needs to be filled in by w1z to calculate things properly
They're just rehashing what other sites are saying - some of it outdated - so I'm not sure where they're getting their info from
I read a bit swedish, and it was translated badly in google translator.
What it actually says is:
6900 is just around the corner, but sources at NordicHardware says AMD's making slow progress.
If the performance is close and the price is much nicer then it will be a big win for AMD, again.
They just need to keep following this strategy imo. :yepp:
about NH article, taking in count 6950 follow surely the same path of 5850, with 2SIMD disabled, less clock, AMD knowing the 570 is around the corner,
if the card is ready and perform enough well for AMD in this segment ( 570 = +- 480 perf ), i don't really see why a 6970 with 2 SIMD more + faster clock is not enough. Even if the card is just 5% under the GTX580,
it's not really a problem ..
real reason why you post such a comment? die hard NV fanboy ????
no performance data known, no price range known, no power consumption known only few weeks delay yet you compare it with a product that was late more then 6 months, consumed way to mcch and was underperforming like hell.... yeah real die hard fanboy. :ROTF:
Barts might be a cheap replacement for 5770. A safe chip where only some of Caymans elements were tested, made to be out as soon as possible to battle GTX 460. Cayman being the flagship have a larger budget and a bit longer development time, and most importantly higher goals.
I think GTX 460 gave AMD a headache, they could of course kill it off with 5870 at the same price level and still earn cash for every sold card. But that would undermine the market for 6800. GTX 460 was way to late to change any of AMDs design plans with 6800, but I think AMD knew that they needed the 6800 long before that to fill the gap in the most important price range.
Dave Baumann, AMD, on B3DQuote:
Barts did not exist on the roadmap in 32nm. Barts in 40nm turned up before the 32nm cancellation.
I wouldn't put it that way. They say there are problems with the 6970 but they don't specify it further than that but they also say that when the card launches the availability will be bad. They also state the obvious by saying Cayman won't launch until 6970 is ready... Somehow though, the 6950 seems to be trickling its way to partners. As I said earlier, NH hasn't been anywhere near accurate anything regarding AMD launches the past year or two.
Do you feel lucky...punk?
http://lab501.ro/wp-content/uploads/...feellucky2.jpg
Well, their Barts stories were quite off. All of them. Also their initial RV870 rumours were off too (they were quite sure about the 2000SPs at first (after RV770 launch) then changed to 1200 before settling on 1600 not until August 2009).
I have tried to google for misinformated posts but I just realized that a lot of their articles have been pulled when their info have been criticized by other respectable sites / forum members.
Monstru:
Are those Cayman?!
Are those 6970's? If so, why so Accordion?
Well 6870 doesn't have a backplate so at least that's new.